Voice Reference: Oath-Bound Content System

In most places, a person’s word matters because other people remember it. Reputation travels, trust builds or breaks, and a man who behaves poorly or speaks rashly finds doors closing to him. That is familiar ground. What is different in Oath-Bound is that a spoken promise is not only remembered — it is almost tangible.

When someone gives their word here, they are not just stating intent. They are fixing a line that others can see, measure against, and act upon. Even the simplest promise begins to shape expectations: who can rely on you, who can challenge you, and what others are entitled to assume you will do. People listen closely, not only to what is said but to what is left unsaid, because both begin to carry weight once spoken aloud. And they watch what you do — because in a world where words carry this kind of weight, deeds are the evidence against which words are tested. A man who speaks well and acts poorly is understood to be something specific, and that understanding travels.

For the most part, a promise made in Oath-Bound carries no more intrinsic force than it would anywhere else. What makes the difference is cultural: most people here understand, at some level, that a promise is a small oath, and everyone knows that oaths have consequences. This awareness shapes how commitments are made, how they are heard, and how seriously their breach is taken. A warlord’s oath to his liege, a merchant’s bond to his trading partner, a sworn declaration before witnesses — these are real and binding in every practical sense, enforced through memory, reputation, and the social consequences of breaking what was given freely. The gods are not involved, and do not need to be.

Some oaths, however, are something more than social arrangements. What people build between themselves can reflect something that exists beyond them — a structure of obligation that runs through the world at every level. The old formulation captures it precisely: as above, so below. The laws that govern oaths among mortals are not an approximation of divine law — they are the same law, differently expressed.

A Bound Oath is one that has been recognized and answered for by the gods themselves. The god becomes a party to the commitment, and what is fixed between the people present is fixed in something larger than either of them. How this happens varies: in more settled societies, a priest mediates the process through formal ritual, ensuring the oath is properly framed and the divine attention properly engaged. In older or less formal traditions, the same end is approached differently — through spectacle, ceremony, and the presence of witnesses enough to make the moment undeniable, often with some object serving as a focus for what is being sworn. Whether such objects accrue significance of their own over time is a matter on which reasonable people in Oath-Bound disagree, though the families and clans that keep them tend to treat the question as settled.

Bound Oaths have shaped the history within which the players live, and the consequences are not difficult to read. About 650 years ago, a group of exiles bound itself with what was probably the first fully mediated Bound Oath in recorded history — a commitment to carve out a new homeland and to govern it in a manner worthy of what had been sworn. The Empire that followed stood for three centuries. The oath was broken. The Empire fell 353 years ago — the campaign present is 353 NA — and the world the players move through is still, in many respects, living inside that consequence. The ruins are present and obvious. The successor states have memories, and the people have memories that are not old enough to have become legends. Even the lowest serf in the most remote corner of Oath-Bound understands on some level that the gods took notice when that oath was made — and that they noticed when it was broken.

For you as a player, most of what you encounter will be obligation of the ordinary kind — promises, debts, sworn loyalties, and the reputational consequences of how you handle them. But Bound Oaths exist, some of them still active, and the circumstances that lead to them are not always formal or anticipated. Understanding that there is a difference between a commitment witnessed by people and one witnessed by something that does not forget is, in Oath-Bound, considered basic literacy.

The campaign present is 353 NA (656 AF). The Fall of the Velasian Empire occurred 353 years ago. No living person has unmediated memory of the Empire. Family traditions that claim to trace back to imperial service are eight to ten generations removed from the events they describe.

Use this anchor when writing content that references elapsed time:

Player characters are typically in their twenties to mid-forties, born in the 307–352 NA range. They are approximately 16 generations removed from the Fall.

This work generates content for four sites with distinct audiences. Apply the correct register to every document before drafting.

These apply across all sites unless a specific site register overrides.

Subjunctive constructions are particularly useful because they can subordinate a condition, concession, or consequence into the main clause without adding length — "a session worth having for most of the people at the table, even where not for all of them" does more work than two sentences would. Chained conjunctions are similarly acceptable where multiple related ideas need to be held together: and, but, or, while, even where, even when, whether or not, and similar.

Apply this test to every paragraph during any polish pass. The default is conjunction or subordination; the staccato form must justify itself by the specific work the pause is doing.

The standing prohibited list:

The standing list is not exhaustive. Any word that describes effect rather than mechanism, or gestures at a claim without making it, is a candidate for replacement, whether or not it appears above.

The diagnostic test applies on every polish pass, including to words not on the standing list. When a word from the slop category is caught in output — by the author or by review — it is added to the list with its context and a confirmed replacement, so the list grows from real examples rather than theoretical ones. Iteration is the method: apply the test, catch the instance, add the entry, tighten the list.

The rationale for this iterative approach is worth stating explicitly. As the corpus becomes more precise, imprecision becomes more visible — a vague word that might pass unnoticed in undifferentiated prose stands out as an island against a surrounding baseline of specific, precise language. The more rigorously the diagnostic is applied, the more obvious each remaining lapse becomes, and the more reliably such lapses function as indicia of AI output rather than authorial choice. Precision compounds: every pass raises the standard against which the next pass is measured.

Content that is identifiably AI in origin is acceptable. Content recognizable as AI slop is not, and the diagnostic test is the primary instrument for telling them apart.

Telegraphing survives only where the point being flagged is genuinely counterintuitive, where the reader might otherwise skim past it, or where the writer is explicitly stepping outside the main argument and the signal is load-bearing. The test: would removing the telegraphing phrase change what the sentence means or how the reader receives it? If not, remove it and let the sentence lead with its content.

"Telegraphing" is the standing term for this category across all Oath-Bound corpus directives.

These rulings apply to all new content and to documents being actively worked on in the current session. They are not applied retroactively to uploaded corpus documents during normal working sessions — back-fixes to existing documents are handled as dedicated cleanup passes.

When a non-conforming term appears in an uploaded document that is being used as source material, note the discrepancy internally and apply the correct form in any output derived from it. Do not interrupt the working session to correct the source document unless a cleanup pass has been explicitly requested.

These rulings exist because the same errors recur across sessions and the cost of propagating them through the corpus is high.

Vale (vale.sh) is the prose linting tool for the Oath-Bound corpus. It is installed via Scoop on Windows (scoop install vale), Homebrew on macOS (brew install vale), or package manager on Linux.

Vale requires Asciidoctor for native AsciiDoc parsing: